You just saved me life. Brought me back from the edge of a ledge. And back to the human race with VARY REGION which is REPAINT in Alpha which is what I needed because I generated the image in Alpha and it's a professional headshot so no creative upscales just stark realism. Thank you 🙏
I like the clear step by step explanations in this tutorial. Focussing on one image and working to achieve your creative vision with the tools and working around their limitations was a great way to structure the video. Nice work.
TOPIC REQUEST: Now that you have this tech-human character, how can you create more pictures that use this same character in different poses and environments? Can you do a tutorial on character consistency across different environments, poses, actions, etc.? Thanks!
Can you show us how to do different angles of the same object? So if I want to do a car, can I get multiple different angles of the same car? Loved this tutorial dude. Subbed!
Very intersting video thx. I would interested into some explanation on how to create prompt with a specific artiste or mangaka like Toriyama/ Kishimoto or others but using a photo of your self as model. It would be like the artiste drew you. +1 sub :)
Is it now possible with some AI tool to blend a picture of a product of a certain design , for example some jewelry and a picture of a person? And then put that product on the person? Something like replacement for real photography set
I have a question: the quality of the image keeps getting worse with each Vary (Region). In other words, after a couple of uses of Vary (Region) editor, the images get a lot darker and lose on quality. Did you experience this? Is there a workaround? I have image references if you need me to clarify.
Great and helpful video! The music track in the background is very soothing (almost meditative) and complements your voice beautifully. Can you share the name of the track, please?
🔥 thanks for putting this together, super helpful. Any chance you make a video on non-charecter image generation? 2d logos or frame styles that could be exported as dxf?
I'm currently experimenting with it for the next video! Getting some really decent consistency but the image quality is awful. Hoping to improve that...
Question: when using the vary region it removes the object (like content aware fill) but it does not replace it with the new object i've prompted. I've also tried --stylize 1000 --chaos 25
Thank your for your response! The original prompt was something like: office guy behind desk. - sref (media link to another image) Then I upscaled it and tried vary region. Here I masked part of the desk where a laptop was standing and prompted: law books and law hammer. All the results where mostly an empty desk on that part with nothing replaced. Sometimes a tiny magazine maybe was recognizable. I also tried -stylize 1000 -chaos 100 but nothing changed much. I will give it another try with another themes, but your thoughts are of course appreciated :)
I found a similar struggle. I wanted to add a katana on the back of the character in this video. Selected the area behind my char and prompted "katana, katana on back" but it didn't add it to the image. I too tried --stylise 1000 and --chaos 100 which didn't help much. What I found to be more helpful was selecting a larger area and giving a more general prompt. I selected the entire body of my character and the area behind it & prompted "samurai with katana on back". That gave me much better results but image still looked a little odd. Hope this helps?
Thanks for this great video! I would like a video about adding a real life image to an AI generated image - f.ex. taking a real image of clothes and putting an AI generated person in those clothes.
![Create Perfect Images - How to use MidJourney Inpaint [AI Art Tutorial 2024]](/img/logo.svg){kind=logo}
